A Single Malicious Webpage Let Researchers Hijack Meta Muse and Access Private Chats
Key Points:
- Researchers demonstrated a malicious webpage could extract private conversations and stored memories from Meta Muse users without triggering approval prompts, potentially accessing connected account credentials.
- The attack was a controlled laboratory demonstration confined to an individual user's cloud machine and did not affect other users or shared servers.
- The exploit involved chaining a flaw in Muse's built-in browser with a Linux kernel vulnerability, enabling full control over the user's cloud environment, including access to credential stores and security layers.
- Muse's security design, which assumes the agent may be compromised and isolates damage, functioned as intended, but the attack still succeeded in breaching the user's machine.
- The security firm NebuSec reported the vulnerabilities to Meta's Muse team, who have since fixed the issues before the research was published; Meta has not publicly confirmed or commented on these fixes.